The incidence of esophageal cancer (EC) is rising at an unprecedented rate. Patient prognosis is related to lymph node (LN) involvement, but disease spread to three distinct LN fields makes accurate nodal assessment difficult. Conventional imaging techniques are inaccurate for LN staging. Minimally invasive LN sampling with histological examination is currently the most accurate staging protocol, but even with radical LN dissection, patients with histologically negative LN have a 30-50 percent chance of developing recurrent disease. This suggests that histologic examination misses microscopically occult metastases (MOM). Our group and others have recently shown that molecular detection of MOM using reverse transcriptase-polymerase chain reaction (RT-PCR) correlates with disease recurrence, but the clinical relevance of these findings is debatable due to a high false positive rate. Our preliminary data shows that quantitative RT-PCR (QRT-PCR) can discriminate between true, and false positives, and thus can accurately identify histologically LN negative patients who have an increased risk of recurrence. We propose to use QRT-PCR to detect MOM in our large population of EC patients, and compare QRT-PCR with conventional histology. In Specific Aim 1 we will measure mRNA expression levels of molecular markers carcinoembryonic antigen (CEA) and cytokeratin 19(CK19) in LN's from EC and benign disease patients. With this information, we will determine the optimal expression level cut-off values for defining positive LN's. We will then compare sensitivity and specificity for recurrence prediction using both QRT-PCR and histologic analysis of LN's. Patients diagnosed with histologically N1 disease are offered adjuvant chemoradiotherapy. Up to 50 percent may have a pathologic complete response but of these, over 50 percent will recur. In Specific Aim 2 we will determine the clinical significance of QRT-PCR positivity in LN's of histologically downstaged patients who have received neoadjuvant treatment for N1 EC. Finally, intraoperative LN staging by histology of frozen sections does not detect MOM, and is incorrect 7-10 percent of the time compared with fixed tissue histology. In Specific Aim 3 we will explore the use of new developments in QRT-PCR technology (SmartCycler), to provide rapid, intraoperative QRT-PCR analysis of LN's and improve the accuracy of surgical staging. This investigation will define the role of QRT-PCR in staging LN's from patients with EC, determine if clinical recurrences can be predicted in patients who were pathologically downstaged by neoadjuvant therapy and assess results and clinical applications of intraoperative QRT-PCR.